/* * Copyright (c) 2017 Cisco and/or its affiliates. * Licensed under the Apache License, Version 2.0 (the "License"); * you may not use this file except in compliance with the License. * You may obtain a copy of the License at: * * http://www.apache.org/licenses/LICENSE-2.0 * * Unless required by applicable law or agreed to in writing, software * distributed under the License is distributed on an "AS IS" BASIS, * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. * See the License for the specific language governing permissions and * limitations under the License. */ #include <math.h> #include <vlib/vlib.h> #include <vnet/vnet.h> #include <vppinfra/elog.h> #include <vnet/session/transport.h> #include <vnet/session/application.h> #include <vnet/session/session_debug.h> #include <svm/queue.h> vlib_node_registration_t session_queue_node; typedef struct { u32 session_index; u32 server_thread_index; } session_queue_trace_t; /* packet trace format function */ static u8 * format_session_queue_trace (u8 * s, va_list * args) { CLIB_UNUSED (vlib_main_t * vm) = va_arg (*args, vlib_main_t *); CLIB_UNUSED (vlib_node_t * node) = va_arg (*args, vlib_node_t *); session_queue_trace_t *t = va_arg (*args, session_queue_trace_t *); s = format (s, "SESSION_QUEUE: session index %d, server thread index %d", t->session_index, t->server_thread_index); return s; } vlib_node_registration_t session_queue_node; #define foreach_session_queue_error \ _(TX, "Packets transmitted") \ _(TIMER, "Timer events") \ _(NO_BUFFER, "Out of buffers") typedef enum { #define _(sym,str) SESSION_QUEUE_ERROR_##sym, foreach_session_queue_error #undef _ SESSION_QUEUE_N_ERROR, } session_queue_error_t; static char *session_queue_error_strings[] = { #define _(sym,string) string, foreach_session_queue_error #undef _ }; always_inline void session_tx_fifo_chain_tail (session_manager_main_t * smm, vlib_main_t * vm, u8 thread_index, svm_fifo_t * fifo, vlib_buffer_t * b0, u32 bi0, u8 n_bufs_per_seg, u32 left_from_seg, u32 * left_to_snd0, u16 * n_bufs, u32 * tx_offset, u16 deq_per_buf, u8 peek_data) { vlib_buffer_t *chain_b0, *prev_b0; u32 chain_bi0, to_deq; u16 len_to_deq0, n_bytes_read; u8 *data0, j; b0->flags |= VLIB_BUFFER_TOTAL_LENGTH_VALID; b0->total_length_not_including_first_buffer = 0; chain_bi0 = bi0; chain_b0 = b0; to_deq = left_from_seg; for (j = 1; j < n_bufs_per_seg; j++) { prev_b0 = chain_b0; len_to_deq0 = clib_min (to_deq, deq_per_buf); *n_bufs -= 1; chain_bi0 = smm->tx_buffers[thread_index][*n_bufs]; _vec_len (smm->tx_buffers[thread_index]) = *n_bufs; chain_b0 = vlib_get_buffer (vm, chain_bi0); chain_b0->current_data = 0; data0 = vlib_buffer_get_current (chain_b0); if (peek_data) { n_bytes_read = svm_fifo_peek (fifo, *tx_offset, len_to_deq0, data0); *tx_offset += n_bytes_read; } else { n_bytes_read = svm_fifo_dequeue_nowait (fifo, len_to_deq0, data0); } ASSERT (n_bytes_read == len_to_deq0); chain_b0->current_length = n_bytes_read; b0->total_length_not_including_first_buffer += chain_b0->current_length; /* update previous buffer */ prev_b0->next_buffer = chain_bi0; prev_b0->flags |= VLIB_BUFFER_NEXT_PRESENT; /* update current buffer */ chain_b0->next_buffer = 0; to_deq -= n_bytes_read; if (to_deq == 0) break; } ASSERT (to_deq == 0 && b0->total_length_not_including_first_buffer == left_from_seg); *left_to_snd0 -= left_from_seg; } always_inline int session_tx_fifo_read_and_snd_i (vlib_main_t * vm, vlib_node_runtime_t * node, session_manager_main_t * smm, session_fifo_event_t * e0, stream_session_t * s0, u32 thread_index, int *n_tx_packets, u8 peek_data) { u32 n_trace = vlib_get_trace_count (vm, node); u32 left_to_snd0, max_len_to_snd0, len_to_deq0, snd_space0; u32 n_bufs_per_evt, n_frames_per_evt, n_bufs_per_frame; transport_connection_t *tc0; transport_proto_vft_t *transport_vft; transport_proto_t tp; u32 next_index, next0, *to_next, n_left_to_next, bi0; vlib_buffer_t *b0; u32 tx_offset = 0, max_dequeue0, n_bytes_per_seg, left_for_seg; u16 snd_mss0, n_bufs_per_seg, n_bufs; u8 *data0; int i, n_bytes_read; u32 n_bytes_per_buf, deq_per_buf, deq_per_first_buf; u32 buffers_allocated, buffers_allocated_this_call; next_index = next0 = smm->session_type_to_next[s0->session_type]; tp = session_get_transport_proto (s0); transport_vft = transport_protocol_get_vft (tp); tc0 = transport_vft->get_connection (s0->connection_index, thread_index); /* Make sure we have space to send and there's something to dequeue */ snd_mss0 = transport_vft->send_mss (tc0); snd_space0 = transport_vft->send_space (tc0); /* Can't make any progress */ if (snd_space0 == 0 || snd_mss0 == 0) { vec_add1 (smm->pending_event_vector[thread_index], *e0); return 0; } /* Allow enqueuing of a new event */ svm_fifo_unset_event (s0->server_tx_fifo); /* Check how much we can pull. */ max_dequeue0 = svm_fifo_max_dequeue (s0->server_tx_fifo); if (peek_data) { /* Offset in rx fifo from where to peek data */ tx_offset = transport_vft->tx_fifo_offset (tc0); if (PREDICT_FALSE (tx_offset >= max_dequeue0)) max_dequeue0 = 0; else max_dequeue0 -= tx_offset; } /* Nothing to read return */ if (max_dequeue0 == 0) return 0; /* Ensure we're not writing more than transport window allows */ if (max_dequeue0 < snd_space0) { /* Constrained by tx queue. Try to send only fully formed segments */ max_len_to_snd0 = (max_dequeue0 > snd_mss0) ? max_dequeue0 - max_dequeue0 % snd_mss0 : max_dequeue0; /* TODO Nagle ? */ } else { /* Expectation is that snd_space0 is already a multiple of snd_mss */ max_len_to_snd0 = snd_space0; } n_bytes_per_buf = vlib_buffer_free_list_buffer_size (vm, VLIB_BUFFER_DEFAULT_FREE_LIST_INDEX); ASSERT (n_bytes_per_buf > MAX_HDRS_LEN); n_bytes_per_seg = MAX_HDRS_LEN + snd_mss0; n_bufs_per_seg = ceil ((double) n_bytes_per_seg / n_bytes_per_buf); n_bufs_per_evt = ceil ((double) max_len_to_snd0 / n_bytes_per_seg); n_frames_per_evt = ceil ((double) n_bufs_per_evt / VLIB_FRAME_SIZE); n_bufs_per_frame = n_bufs_per_seg * VLIB_FRAME_SIZE; deq_per_buf = clib_min (snd_mss0, n_bytes_per_buf); deq_per_first_buf = clib_min (snd_mss0, n_bytes_per_buf - MAX_HDRS_LEN); n_bufs = vec_len (smm->tx_buffers[thread_index]); left_to_snd0 = max_len_to_snd0; for (i = 0; i < n_frames_per_evt; i++) { /* Make sure we have at least one full frame of buffers ready */ if (PREDICT_FALSE (n_bufs < n_bufs_per_frame)) { vec_validate (smm->tx_buffers[thread_index], n_bufs + n_bufs_per_frame - 1); buffers_allocated = 0; do { buffers_allocated_this_call = vlib_buffer_alloc (vm, &smm->tx_buffers [thread_index] [n_bufs + buffers_allocated], n_bufs_per_frame - buffers_allocated); buffers_allocated += buffers_allocated_this_call; } while (buffers_allocated_this_call > 0 && ((buffers_allocated + n_bufs < n_bufs_per_frame))); n_bufs += buffers_allocated; _vec_len (smm->tx_buffers[thread_index]) = n_bufs; if (PREDICT_FALSE (n_bufs < n_bufs_per_frame)) { vec_add1 (smm->pending_event_vector[thread_index], *e0); return -1; } ASSERT (n_bufs >= n_bufs_per_frame); } vlib_get_next_frame (vm, node, next_index, to_next, n_left_to_next); while (left_to_snd0 && n_left_to_next) { /* * Handle first buffer in chain separately */ /* Get free buffer */ ASSERT (n_bufs >= 1); bi0 = smm->tx_buffers[thread_index][--n_bufs]; _vec_len (smm->tx_buffers[thread_index]) = n_bufs; /* usual speculation, or the enqueue_x1 macro will barf */ to_next[0] = bi0; to_next += 1; n_left_to_next -= 1; b0 = vlib_get_buffer (vm, bi0); b0->error = 0; b0->flags |= VNET_BUFFER_F_LOCALLY_ORIGINATED; b0->current_data = 0; b0->total_length_not_including_first_buffer = 0; len_to_deq0 = clib_min (left_to_snd0, deq_per_first_buf); data0 = vlib_buffer_make_headroom (b0, MAX_HDRS_LEN); if (peek_data) { n_bytes_read = svm_fifo_peek (s0->server_tx_fifo, tx_offset, len_to_deq0, data0); if (n_bytes_read <= 0) goto dequeue_fail; /* Keep track of progress locally, transport is also supposed to * increment it independently when pushing the header */ tx_offset += n_bytes_read; } else { n_bytes_read = svm_fifo_dequeue_nowait (s0->server_tx_fifo, len_to_deq0, data0); if (n_bytes_read <= 0) goto dequeue_fail; } b0->current_length = n_bytes_read; left_to_snd0 -= n_bytes_read; *n_tx_packets = *n_tx_packets + 1; /* * Fill in the remaining buffers in the chain, if any */ if (PREDICT_FALSE (n_bufs_per_seg > 1 && left_to_snd0)) { left_for_seg = clib_min (snd_mss0 - n_bytes_read, left_to_snd0); session_tx_fifo_chain_tail (smm, vm, thread_index, s0->server_tx_fifo, b0, bi0, n_bufs_per_seg, left_for_seg, &left_to_snd0, &n_bufs, &tx_offset, deq_per_buf, peek_data); } /* Ask transport to push header after current_length and * total_length_not_including_first_buffer are updated */ transport_vft->push_header (tc0, b0); /* *INDENT-OFF* */ SESSION_EVT_DBG(SESSION_EVT_DEQ, s0, ({ ed->data[0] = e0->event_type; ed->data[1] = max_dequeue0; ed->data[2] = len_to_deq0; ed->data[3] = left_to_snd0; })); /* *INDENT-ON* */ VLIB_BUFFER_TRACE_TRAJECTORY_INIT (b0); if (PREDICT_FALSE (n_trace > 0)) { session_queue_trace_t *t0; vlib_trace_buffer (vm, node, next_index, b0, 1 /* follow_chain */ ); vlib_set_trace_count (vm, node, --n_trace); t0 = vlib_add_trace (vm, node, b0, sizeof (*t0)); t0->session_index = s0->session_index; t0->server_thread_index = s0->thread_index; } vlib_validate_buffer_enqueue_x1 (vm, node, next_index, to_next, n_left_to_next, bi0, next0); } vlib_put_next_frame (vm, node, next_index, n_left_to_next); } /* If we couldn't dequeue all bytes mark as partially read */ if (max_len_to_snd0 < max_dequeue0) { /* If we don't already have new event */ if (svm_fifo_set_event (s0->server_tx_fifo)) { vec_add1 (smm->pending_event_vector[thread_index], *e0); } } return 0; dequeue_fail: /* * Can't read from fifo. If we don't already have an event, save as partially * read, return buff to free list and return */ clib_warning ("dequeue fail"); if (svm_fifo_set_event (s0->server_tx_fifo)) { vec_add1 (smm->pending_event_vector[thread_index], *e0); } vlib_put_next_frame (vm, node, next_index, n_left_to_next + 1); _vec_len (smm->tx_buffers[thread_index]) += 1; return 0; } int session_tx_fifo_peek_and_snd (vlib_main_t * vm, vlib_node_runtime_t * node, session_manager_main_t * smm, session_fifo_event_t * e0, stream_session_t * s0, u32 thread_index, int *n_tx_pkts) { return session_tx_fifo_read_and_snd_i (vm, node, smm, e0, s0, thread_index, n_tx_pkts, 1); } int session_tx_fifo_dequeue_and_snd (vlib_main_t * vm, vlib_node_runtime_t * node, session_manager_main_t * smm, session_fifo_event_t * e0, stream_session_t * s0, u32 thread_index, int *n_tx_pkts) { return session_tx_fifo_read_and_snd_i (vm, node, smm, e0, s0, thread_index, n_tx_pkts, 0); } always_inline stream_session_t * session_event_get_session (session_fifo_event_t * e, u8 thread_index) { return session_get_if_valid (e->fifo->master_session_index, thread_index); } void dump_thread_0_event_queue (void) { session_manager_main_t *smm = vnet_get_session_manager_main (); vlib_main_t *vm = &vlib_global_main; u32 my_thread_index = vm->thread_index; session_fifo_event_t _e, *e = &_e; stream_session_t *s0; int i, index; i8 *headp; svm_queue_t *q; q = smm->vpp_event_queues[my_thread_index]; index = q->head; for (i = 0; i < q->cursize; i++) { headp = (i8 *) (&q->data[0] + q->elsize * index); clib_memcpy (e, headp, q->elsize); switch (e->event_type) { case FIFO_EVENT_APP_TX: s0 = session_event_get_session (e, my_thread_index); fformat (stdout, "[%04d] TX session %d\n", i, s0->session_index); break; case FIFO_EVENT_DISCONNECT: s0 = session_get_from_handle (e->session_handle); fformat (stdout, "[%04d] disconnect session %d\n", i, s0->session_index); break; case FIFO_EVENT_BUILTIN_RX: s0 = session_event_get_session (e, my_thread_index); fformat (stdout, "[%04d] builtin_rx %d\n", i, s0->session_index); break; case FIFO_EVENT_RPC: fformat (stdout, "[%04d] RPC call %llx with %llx\n", i, (u64) (e->rpc_args.fp), (u64) (e->rpc_args.arg)); break; default: fformat (stdout, "[%04d] unhandled event type %d\n", i, e->event_type); break; } index++; if (index == q->maxsize) index = 0; } } static u8 session_node_cmp_event (session_fifo_event_t * e, svm_fifo_t * f) { stream_session_t *s; switch (e->event_type) { case FIFO_EVENT_APP_RX: case FIFO_EVENT_APP_TX: case FIFO_EVENT_BUILTIN_RX: if (e->fifo == f) return 1; break; case FIFO_EVENT_DISCONNECT: break; case FIFO_EVENT_RPC: s = session_get_from_handle (e->session_handle); if (!s) { clib_warning ("session has event but doesn't exist!"); break; } if (s->server_rx_fifo == f || s->server_tx_fifo == f) return 1; break; default: break; } return 0; } u8 session_node_lookup_fifo_event (svm_fifo_t * f, session_fifo_event_t * e) { session_manager_main_t *smm = vnet_get_session_manager_main (); svm_queue_t *q; session_fifo_event_t *pending_event_vector, *evt; int i, index, found = 0; i8 *headp; u8 thread_index; ASSERT (e); thread_index = f->master_thread_index; /* * Search evt queue */ q = smm->vpp_event_queues[thread_index]; index = q->head; for (i = 0; i < q->cursize; i++) { headp = (i8 *) (&q->data[0] + q->elsize * index); clib_memcpy (e, headp, q->elsize); found = session_node_cmp_event (e, f); if (found) break; if (++index == q->maxsize) index = 0; } /* * Search pending events vector */ pending_event_vector = smm->pending_event_vector[thread_index]; vec_foreach (evt, pending_event_vector) { found = session_node_cmp_event (evt, f); if (found) { clib_memcpy (e, evt, sizeof (*evt)); break; } } return found; } static uword session_queue_node_fn (vlib_main_t * vm, vlib_node_runtime_t * node, vlib_frame_t * frame) { session_manager_main_t *smm = vnet_get_session_manager_main (); session_fifo_event_t *my_pending_event_vector, *pending_disconnects, *e; session_fifo_event_t *my_fifo_events; u32 n_to_dequeue, n_events; svm_queue_t *q; application_t *app; int n_tx_packets = 0; u32 my_thread_index = vm->thread_index; int i, rv; f64 now = vlib_time_now (vm); void (*fp) (void *); SESSION_EVT_DBG (SESSION_EVT_POLL_GAP_TRACK, smm, my_thread_index); /* * Update transport time */ transport_update_time (now, my_thread_index); /* * Get vpp queue events */ q = smm->vpp_event_queues[my_thread_index]; if (PREDICT_FALSE (q == 0)) return 0; my_fifo_events = smm->free_event_vector[my_thread_index]; /* min number of events we can dequeue without blocking */ n_to_dequeue = q->cursize; my_pending_event_vector = smm->pending_event_vector[my_thread_index]; pending_disconnects = smm->pending_disconnects[my_thread_index]; if (!n_to_dequeue && !vec_len (my_pending_event_vector) && !vec_len (pending_disconnects)) return 0; SESSION_EVT_DBG (SESSION_EVT_DEQ_NODE, 0); /* * If we didn't manage to process previous events try going * over them again without dequeuing new ones. */ /* XXX: Block senders to sessions that can't keep up */ if (0 && vec_len (my_pending_event_vector) >= 100) { clib_warning ("too many fifo events unsolved"); goto skip_dequeue; } /* See you in the next life, don't be late */ if (pthread_mutex_trylock (&q->mutex)) return 0; for (i = 0; i < n_to_dequeue; i++) { vec_add2 (my_fifo_events, e, 1); svm_queue_sub_raw (q, (u8 *) e); } /* The other side of the connection is not polling */ if (q->cursize < (q->maxsize / 8)) (void) pthread_cond_broadcast (&q->condvar); pthread_mutex_unlock (&q->mutex); vec_append (my_fifo_events, my_pending_event_vector); vec_append (my_fifo_events, smm->pending_disconnects[my_thread_index]); _vec_len (my_pending_event_vector) = 0; smm->pending_event_vector[my_thread_index] = my_pending_event_vector; _vec_len (smm->pending_disconnects[my_thread_index]) = 0; skip_dequeue: n_events = vec_len (my_fifo_events); for (i = 0; i < n_events; i++) { stream_session_t *s0; /* $$$ prefetch 1 ahead maybe */ session_fifo_event_t *e0; e0 = &my_fifo_events[i]; switch (e0->event_type) { case FIFO_EVENT_APP_TX: s0 = session_event_get_session (e0, my_thread_index); if (PREDICT_FALSE (!s0)) { clib_warning ("It's dead, Jim!"); continue; } /* Can retransmit for closed sessions but can't do anything if * session is not ready or closed */ if (PREDICT_FALSE (s0->session_state < SESSION_STATE_READY)) { vec_add1 (smm->pending_event_vector[my_thread_index], *e0); continue; } /* Spray packets in per session type frames, since they go to * different nodes */ rv = (smm->session_tx_fns[s0->session_type]) (vm, node, smm, e0, s0, my_thread_index, &n_tx_packets); /* Out of buffers */ if (PREDICT_FALSE (rv < 0)) { vlib_node_increment_counter (vm, node->node_index, SESSION_QUEUE_ERROR_NO_BUFFER, 1); continue; } break; case FIFO_EVENT_DISCONNECT: /* Make sure disconnects run after the pending list is drained */ if (!e0->postponed) { e0->postponed = 1; vec_add1 (smm->pending_disconnects[my_thread_index], *e0); continue; } s0 = session_get_from_handle (e0->session_handle); stream_session_disconnect (s0); break; case FIFO_EVENT_BUILTIN_RX: s0 = session_event_get_session (e0, my_thread_index); if (PREDICT_FALSE (!s0)) continue; svm_fifo_unset_event (s0->server_rx_fifo); app = application_get (s0->app_index); app->cb_fns.builtin_server_rx_callback (s0); break; case FIFO_EVENT_RPC: fp = e0->rpc_args.fp; (*fp) (e0->rpc_args.arg); break; default: clib_warning ("unhandled event type %d", e0->event_type); } } _vec_len (my_fifo_events) = 0; smm->free_event_vector[my_thread_index] = my_fifo_events; vlib_node_increment_counter (vm, session_queue_node.index, SESSION_QUEUE_ERROR_TX, n_tx_packets); SESSION_EVT_DBG (SESSION_EVT_DEQ_NODE, 1); return n_tx_packets; } /* *INDENT-OFF* */ VLIB_REGISTER_NODE (session_queue_node) = { .function = session_queue_node_fn, .name = "session-queue", .format_trace = format_session_queue_trace, .type = VLIB_NODE_TYPE_INPUT, .n_errors = ARRAY_LEN (session_queue_error_strings), .error_strings = session_queue_error_strings, .state = VLIB_NODE_STATE_DISABLED, }; /* *INDENT-ON* */ static clib_error_t * session_queue_exit (vlib_main_t * vm) { if (vec_len (vlib_mains) < 2) return 0; /* * Shut off (especially) worker-thread session nodes. * Otherwise, vpp can crash as the main thread unmaps the * API segment. */ vlib_worker_thread_barrier_sync (vm); session_node_enable_disable (0 /* is_enable */ ); vlib_worker_thread_barrier_release (vm); return 0; } VLIB_MAIN_LOOP_EXIT_FUNCTION (session_queue_exit); /* * fd.io coding-style-patch-verification: ON * * Local Variables: * eval: (c-set-style "gnu") * End: */